Why track legal document changes?
Companies routinely update their Terms of Service and Privacy Policies — often with minimal notice. These changes can have significant legal implications: adding mandatory arbitration clauses, expanding data collection scope, introducing class action waivers, or modifying liability limitations. By archiving and diffing these documents over time, investigators, journalists, and consumers can detect when rights were quietly removed or obligations silently expanded.
How does word-level diffing work?
Unlike line-level diffs (like git diff), this tool compares documents at the word level using a longest-common-subsequence algorithm. This means even if a paragraph was reformatted or line-wrapped differently, only the actual word changes are highlighted. Additions appear in green, removals in red strikethrough. The tool also classifies changes by legal significance: arbitration, data sharing, GDPR/CCPA rights, liability, and scope changes each get distinct alerts.
Key Terminology
- Arbitration Clause
- A contract provision requiring disputes to be resolved through private arbitration rather than court proceedings. Often paired with class action waivers, these clauses significantly limit consumer legal remedies.
- Class Action Waiver
- A provision preventing users from joining collective lawsuits. When combined with arbitration, individual users must pursue claims separately — making small-dollar claims economically impractical.
- Data Processing Agreement
- A contract governing how personal data is handled between a data controller and processor, required under GDPR Article 28. Changes to these terms can alter what data is shared and with whom.
- Material Change
- A modification to terms that substantially affects user rights, obligations, or expectations. Many policies require notification for "material changes" but define the term vaguely.
📜 Policy Diff Engine — Frequently Asked Questions
How does the Policy Diff Engine detect silent legal changes?
The tool fetches every unique archived version of a site's legal pages from the Wayback Machine, extracts the text content, and runs a word-level diff algorithm between consecutive versions. Changes are classified by type: additions are green, removals are red, and specific legal keywords like arbitration, class action waiver, data sharing, GDPR, and liability limitation are highlighted with alerts when they first appear or disappear.
What legal pages does this tool check?
It scans 25+ common legal page paths including /terms, /tos, /terms-of-service, /privacy, /privacy-policy, /legal, /acceptable-use, /aup, /cookie-policy, /gdpr, /ccpa, /dmca, /disclaimer, /eula, /data-processing, /sla, and /refund-policy. You can also add custom paths for non-standard URL structures.